-
35
12th International Conference on Fluidized Bed Technology
CYCLONE SYSTEMS IN CIRCULATING FLUIDIZED BEDS
Ted M. Knowlton
Particulate Solid Research, Inc., 4201 W. 36th Street, Chicago,
IL 60632, USA
E-mail: [email protected]
Abstract Cyclones are an integral part of nearly all fluidized
bed processes, and especially so for circulating fluidized bed
(CFB) systems. Cyclones are extremely important to the successful
operation of nearly all CFB processes. The two most important CFB
processes - fluidized catalytic cracking (FCC) and circulating
fluidized bed combustion (CFBC) - operate with different particle
sizes and at different operating conditions. Therefore, the cyclone
designs for each of these major CFB processes are also different.
Cyclones for CFBC units are generally very large (8 to 10 m in
diameter) and typically have only one stage. Cyclones for FCC units
are much smaller (of the order of 1.2 to 2 m in diameter) and are
designed for a minimum of 2 and as many as 4 stages in series. The
average particle size flowing around FCC units is only about 70
microns, while the particles circulating in CFBC units are
typically 150 to 200 microns. In this paper, how the differences in
cyclone operation and design affect CFB system operation is
described and discussed.
INTRODUCTION Using centrifugal force to separate solids from a
gas was first patented in the United States in 1885 by an employee
of the Knicker Bocker company named John Finch (Finch, 1885). Since
then, cyclones have been one of the most important devices in
particulate processes. Cyclones separate solids from a gas stream
at a low energy cost, are relatively inexpensive and because they
have no moving parts are very reliable. In general, cyclones are
used to separate solids from gases in approximately 99% of all
fluidized bed processes. Processes without cyclones either use
filters, inertial separators or in some cases use no solids/gas
separation at all (Burdett et al., 2001; Maryamchik and Wietzke,
2010).
Cyclones are unique devices in that they can be used over
extremely wide solid loading ranges and over extremely large size
ranges. Cyclone sizes range from 1 to 2 cm in diameter for
laboratory units, up to 8- to 10-m diameter cyclones in circulating
fluidized bed combustion (CFBC) units. Cyclone solids loadings can
range between about 0.0002 kg/m3 (0.00013 kgs/kgg) and nearly 50
kg/m3 (42 kgs/kgg). This is a factor of 250,000!
Cyclone operation is not the same over this wide loading range.
For example, secondary cyclones operating in catalytic processes at
low loadings experience erosion in their cones, while at higher
loadings they do not. The dependence of pressure drop with
increasing loading is different depending on whether the cyclone is
operating in the high or low loading regime. In addition, the type
of cyclone inlet (tangential or volute) can greatly influence
efficiency at high loadings, but is not quite so important at low
loadings. Also, the operation of cyclone diplegs is significantly
affected by whether the cyclone is a high-loaded or a low-loaded
cyclone.
Cyclone loadings have traditionally been expressed on a weight
basis in one of two ways: 1) in kg of solids (kgs) /kg of carrying
gas (kgg), or 2) in kg of solids per m
3 of carrying gas. Either one of these two methods is generally
a satisfactory way of expressing and comparing loadings if the
particle densities of the solids are relatively close. However, if
the solids differ widely in particle density, the concentration of
the solids in the carrying gas can vary significantly when the
loading is expressed on a weight basis (Knowlton and Karri, 2008).
This can be seen in Table 1. This table compares three materials
with different particle densities (600, 1500 and 3600 kg/m3). The
low-density material is representative of a resin, the
middle-density material that of a coal char or a dense FCC
catalyst, and the high-density material is similar to the density
of an ore (iron, titanium, etc.). If each material is added to a
cyclone at the same weight-based loading of 18 kg/m3, the
volumetric loading expressed as solids volume fraction (1- ) varies
from 0.03 (3% solids) for the resin to 0.005 (0.5% solids) for the
ore. Thus, the material with a particle density of 600 kg/m3 has a
solids volume fraction 6 times the solids volume fraction of the
material with a particle density of 3600 kg/m3.
However, because the traditional way of expressing loading is on
a weight or mass basis, loading will be expressed as a mass loading
in this paper recognizing that solids volume fraction or solids
concentration is probably a more realistic way of expressing
loading - especially for comparing materials of widely differing
particle densities.
-
36
Table 1. Comparison of Loadings Based on Volume and Weight
Material Loading,
kg/m3 Particle Density,
kg/m3 Solids Volume Fraction, 1-
1 18 600 0.030 2 18 1500 0.012 3 18 3600 0.005
Cyclones are generally classified as either low-loaded or
high-loaded cyclones. The demarcation between high and low loading
is arbitrary, and, in general, high-loaded cyclones are primary
(first stage) cyclones, and low-loaded cyclones are secondary
(second stage), tertiary (third stage) or even fourth stage
cyclones. Particulate Solid Research, Inc. (PSRI) classifies
cyclones with a loading of greater than 1 kgs/kgg, as high-loaded
cyclones. Cyclones with a loading less than 1 kgs/kgg, are
classified as low-loaded cyclones.
Another way to define low-loaded and high-loaded cyclones is to
specify that cyclones are low-loaded cyclones when their pressure
drop decreases with an increase in solids loading (Fig. 1, Knowlton
and Karri, 2008), and high-loaded cyclones when their pressure drop
increases with solids loading.
The reason that the cyclone pressure drop initially decreases
with increasing loading is that the solids on the cyclone wall
effectively roughen the wall and increase the frictional resistance
to gas flow. This causes the tangential velocity in the cyclone to
decrease, thus decreasing the pressure drop (Yuu, et. al., 1979),
as can be seen in Fig. 2. In this figure, the tangential velocity
decreased from 16 to 13 m/s at the wall, and from about 34 to 17
m/s at its maximum value as solids were added to the cyclone. At
higher loadings, the pressure drop due to the solids acceleration
becomes so large that the total cyclone pressure drop then
increases as solids loading increases.
CFB Systems In commercial CFB systems, the two dominant
technologies using cyclones are the fluidized catalytic cracking
(FCC) and the circulating fluidized bed combustion (CFBC)
processes. There are significant differences in the two processes
that are summarized in Table 2. One of the primary differences is
that FCC processes use Geldart Group A solids, and CFBC processes
use Geldart Group B solids. Another area utilizing CFBs is chemical
looping, with several different types of chemical looping processes
being developed. Many of these processes use Geldart Group A
solids, and so their cyclone design and operation will be similar
to FCC cyclones.
One type of an FCC unit is shown in Fig. 3. In an FCC unit the
circulating fluidized bed is the riser. In the riser, oil is added
to the hot circulating catalyst (approximately 70 microns in size)
at the bottom. The hot catalyst cracks the oil into various
components (kerosene, gasoline, diesel oil, etc.). The catalyst and
the
0.01 0.03 0.1 0.3 1 3 10 30 1000
0.5
1
1.5
2
2.5Cyclone Dia: 250 mm (10 in)Volute Inlet
Material: 76 micron FCC CatalystInlet Gas Velocity, m/s
(ft/s)
19.8 (65)15.2 (50)11.3 (37)
Loading at Cyclone Inlet, kg/m3
Cyc
lone
Pre
ssur
e D
rop,
kPa
Fig. 1. The effect of loading on cyclone pressure drop
Fig. 2. Cyclone tangential velocity with and without solids flow
in the cyclone
0 50 100 1500
10
20
30
40
Cyclone Wall
Gas Outlet Tube Wall
Gas Only
Gas With Solids
Cyclone Radius, mm
Tang
entia
l Vel
ocity
, m/s
Inlet Velocity = 18 m/s
-
37
products are then routed into cyclones at the top of the riser.
Post-cracking of the product gases (which is detrimental) can occur
if the solids and the gases are not separated quickly. Therefore,
the gas and solids are separated as quickly as possible by routing
them through close-coupled cyclones or other inertial separators
(such as an RSS or a VSS vortex separator, etc.). The loading in
the close-coupled cyclones can be very high (up to 24 to 32 kg/m3).
These close-coupled cyclones are some of the highest loading
cyclones in the world, and generally have significantly higher
solid inlet loadings than CFBC cyclones. If close-coupled cyclones
are used, the FCC riser discharges into generally 4 to 6 parallel
cyclones located close to the riser. A second stage of cyclones in
series with the first stage is also used.
Table 2. Comparison of FCC Riser and CFBC Processes (Typical
Values)
Parameter FCC Riser CFBC
Temperature, C 550 850 - 900
Pressure, Bar(g) 2 0.1
Particle Size, microns 70 150 - 200
Geldart Group A B
Riser Solids Flux, kg/s/m2 500 - 1000 20 - 50
Riser Suspension Density, kg/m3 8 to 32 2 to 10
Velocity, m/s 14 - 20 3 to 5
Riser Geometry Circular Rectangular
Riser Diameter, m 0.9 - 1.5 ----------
Boiler Dimensions (large)*, m --------- 28 x 11
Height, m 28 35 48*
*Lagisza CFBC (460 MWe)
As the catalyst flows up the FCC riser, carbon is deposited on
the catalyst, reducing its activity. Therefore, the catalyst is
routed to a fluidized bed regenerator, where the carbon is burned
off of the catalyst using air to
restore its activity. However, if the catalyst is sent to the
regenerator directly after passing through the cyclones, the
valuable product gases in the interstices of the solids would be
burned up. Therefore, the catalyst is passed through a steam
stripper to remove product gases. After most of the product gases
are removed, the catalyst is then routed to the regenerator. The
regenerator is a large unit (usually 10 to 16 m in diameter) and
contains many cyclone stages in parallel. Typically, the number of
parallel stages of cyclones in the regenerator ranges from about 4
for small units to 22 for very large units. Because each primary
cyclone has a secondary cyclone associated with it, for the largest
unit indicated above, 44 cyclones are suspended in the freeboard of
the regenerator. These cyclones can take up as much as 33 to 50% of
the area in the freeboard.
Because particulate emission limits are very low (of the order
of 50 mg/Nm3), a third stage cyclone separator (TSS) is generally
added downstream of the secondary cyclones. Most TSS units are
composed of smaller multiclones with axial inlet vanes in parallel
(Fig. 4, Weaver and Geiger, 2002). In large units, more than one
hundred multiclones
(also called swirl tubes) may be in one TSS shell. To enhance
the performance of the TSS, about 3% of the
Air
Flue Gas
StrippingSteam
Product
Oil Feed
RiserReactor
Regenerator
Slide Valve
Fig. 3. Schematic Drawing of a Side-by-Side FCC Unit
-
38
cyclone flow is pulled out of the bottom of the cyclone with the
solids. This higher loading gas stream is then sent to either a
fourth stage separator (FSS), which is also typically an axial
inlet type of cyclone, or to a filter depending on the emission
requirements.
Important differences in operation, scale and configuration
between FCC unit cyclones and CFBC cyclones are shown in Table 3.
CFBC systems have fewer and larger cyclones than in FCC units
(Table 3). Most of the CFBC cyclones are also external to the
riser, whereas most of the cyclones in an FCC unit are internal
cyclones. One type is not really better than the other. Whether you
use an internal or an external cyclone really depends on what is
better for the process.
A schematic drawing of one type of CFBC unit is shown in Fig. 5.
In this unit, coal is fed into the bottom
of the rectangular CFB riser with limestone, which absorbs much
of the SO2 given off during combustion of the coal. The coal is
combusted in the riser using air, and the solids and the combustion
gases pass into the large cyclone(s). The solids are returned to
the bottom of the riser via a dipleg and, usually, a loop seal. The
CFB combustor units are rectangular and, depending on their size,
have only one large cyclone if relatively small, or 2 to 8 cyclones
in parallel if they are larger units. Only one stage of cyclone (or
solids separator) is typically used in CFBC systems. The particles
entering the cyclones are larger (150 to 200 microns vs. 70) and
denser than in FCC units. Particles in CFBCs have particle
densities ranging from 1800 to 2400 kg/m3 while FCC catalyst
densities are typically 1200 to 1500 kg/m3.
Table 3. Comparison of FCC and CFBC Cyclones
Cyclone Parameter FCC CFBC
Diameter*, m 1.2 2.2 5 - 10
Inlet Velocity, m/s 13.5 - 20 18 - 28
Outlet Velocity, m/s 20 - 25 22 - 38
Number of Stages in series 2 4 1
Number of parallel stages 4 - 22 1-8
Loading, kg/m3 10 35 2 - 10
Operating Temperature, C 550 850 - 950
Average particle size in incoming gas, microns 70 150 - 200
Primary Cyclone Efficiency, % 99.95+ 99.9+
Secondary Cyclone Efficiency, % Up to 98% NA
*Some CFBC cyclones are actually solid separators and can be
rectangular, hexagonal, etc.
Fig. 4. Third-Stage Separator (Shell type) with axial inlet
swirl tubes
SecondaryAir
Air In
Cyclone
BedDischarge
Primary Air
Coal-LimestoneFeed LoopSeal
Standpipe
Riser
Fig 5. Schematic Drawing of a CFBC
-
39
Cyclone Types and Configurations There are several different
cyclone configurations used in CFB systems, and they are generally
classified by inlet type. The two most commonly-used cyclones are
the tangential and volute inlet cyclones (Fig. 6). The volute-inlet
cyclone has been found to be more efficient for highly-loaded
cyclones than the tangential inlet. This is because the tangential
inlet cyclones can produce an interference eddy near the inlet of
the cyclone. This eddy causes fluctuations in the inlet solids
stream that can cause the inlet solids to impact the gas outlet
tube. This results in lower efficiency, and can cause erosion of
the gas outlet tube as well. If a tangential cyclone is used for a
highly-loaded cyclone, the inlet should be located such that the
distance between the cyclone inlet and the wall of the gas outlet
tube is great enough so that the fluctuating solids do not impact
the gas outlet tube. This generally means that this type of
tangential inlet cyclone will have a larger diameter than a typical
tangential inlet cyclone.
A highly-loaded, volute inlet cyclone does not suffer from this
problem. With the volute inlet, the solids have experienced a
significant centrifugal force before entering the cyclone barrel,
enter the cyclone at an angle and, therefore, the entering solids
do not experience the interference from the rotating solids stream
in the cyclone barrel. This difference is depicted in Fig. 7.
Therefore, in almost all highly-loaded cyclones (essentially all
primary cyclones) a volute inlet is used. For low-loaded cyclones,
the loading is so low (approximately 1/1000 or less of that of the
primary cyclone) that the interference with the inlet solids stream
does not occur, so tangential inlet cyclones are satisfactory
designs for low-loaded cyclones. Volute cyclones are equally
satisfactory as well, but quite often tangential cyclones are used
for secondary cyclones because tangential inlet cyclones are less
expensive than volute inlet cyclones.
In CFBC cyclones, the inlet to the cyclone has a slightly
different configuration than for FCC cyclones. The inner wall of
the inlet is angled toward the wall of the cyclone. To keep the
area of the inlet approximately constant, the bottom of the inlet
is angled downward at approximately 30. This type is inlet is
depicted in Fig. 8. This inlet configuration is also used in non-
CFBC cyclones as well, and especially so in Europe. This inlet is a
superior inlet compared to the straight tangential inlet tangential
inlet shown in Fig. 6, because it directs the solids to the wall of
the cyclone where you want them to be for high solids collection
efficiency.
An axial inlet cyclone of the type shown in Fig. 4 uses axial
swirl vanes to impart centrifugal force to the particles. This type
of inlet is generally used for smaller-diameter cyclones, such as
the multiclone TSS units also shown in Fig. 4.
Volute Inlet Tangential Inlet
Fig. 6. Schematic Drawing of Tangential and Volute Cyclone
Inlets
Solids Entering a Tangential Cyclone Expandand Can Impact the
Gas Outlet Tube if theVortex Tube/Cyclone Wall Distance is notGreat
Enough
This is Not the Case for a Volute InletBecause Solids Enter the
Cyclone atan Angle. They Have Also AlreadyExperienced a Centrifugal
Force BeforeEntering the Cyclone
At High Solid LoadingsFluctuating Solids Can
Impinge on theTube if the Wall/Tube
TANGENTIAL INLET
VOLUTE INLET
Solids Do NotImpinge OnGas Outlet ube
Distance is notGreat Enough
Fig. 7. Tangential and Volute Inlet Operation at High Solids
Loading
-
40
Cyclone efficiencies of CFBC primary cyclones can be very high,
99.9%, and nearly rival the efficiencies of the FCC primary
cyclones (even though the cyclones are much larger in diameter and
the centrifugal force on the solids is significantly less with the
larger cyclones). The particles in the CFBC units are both larger
and more dense, which partly explains why the larger-diameter
cyclones are so efficient. However, another factor comes into play
with primary cyclones. The solids loading is generally so high that
the primary cyclone operates like an inertial separator for much of
the solids as explained by Muschelknautz et. al. (1996). In their
model, the gas can carry only a maximum amount of solids (up to
what is called the critical loading). At any solids loading in
excess of this critical loading, the solids are immediately
separated from the gas at the inlet to the cyclone as indicated in
Fig. 9. The solids remaining in the gas are then separated in the
cyclone barrel and primarily in the inner vortex below the gas
outlet tube as if the cyclone were operating at a low solids
loading. However, the inner vortex separation is reduced as the
solids loading increases (Muschelknautz, 2010). Because the solids
at high loading are separated largely by inertial separation, the
efficiencies of a high loaded, large-diameter cyclone can approach
that of a smaller cyclone. This inertial unloading effect has been
noted visually at PSRI in a 53-cm diameter Plexiglas cyclone. The
solids enter the cyclone and then immediately fall down to the
solids outlet. The solids entering high-loaded cyclones have only
about 1.5 turns as they fall down the cyclone from the inlet to the
solids outlet. A low-loaded cyclone was visually observed to
have
approximately 5 to 7 spirals in the same cyclone. High loading
also affects the vortex below the gas outlet tube. The large mass
of solids attenuates, or reduces, the spin rate of the vortex in
high-loaded cyclones. Hoffman et. al., 1995 reported that the
vortex length (swirl intensity) decreased with increasing solids
loading, and was a strong function of the cyclone length as
well.
As mentioned above, FCC and other Group A catalyst processes
have several internal cyclones in parallel. The parallel cyclone
configuration is used to reduce the length of the reactor vessel.
Large-diameter cyclones also are very long cyclones. If a single
internal cyclone were used inside the reactor vessel, the freeboard
above the fluidized bed would be very tall. This would increase the
length of the vessel and result in a tremendous increase in cost
because the vessels are large in diameter. Therefore, several
parallel cyclones are used. Also, a single external cyclone would
return the solids to only one side of the bed, and the solids may
not distribute well from this one feed point throughout the
fluidized bed. The external cyclone would also have to be a
pressure vessel, and insulated much more than for the internal
cyclone. Essentially all internal FCC cyclones have a refractory
hex-mesh
lining, approximately 2.5-cm thick, used primarily to minimize
erosion of the cyclone.
CFBC processes use large external cyclones which (if not
water-cooled) have refractory linings of the order of 300 mm thick.
For cyclones with water-cooled walls, the wall is covered with a
high-conductivity refractory that is about 5 cm thick. The cyclones
are very large in diameter (up to 10 m) and correspondingly very
long. However, the riser furnace is also very tall, and can
accommodate the large, long cyclone as well as the dipleg and loop
seal below it. For the Lagisza CFBC referenced in Table 2, the
height of the riser is 48 m. External cyclones have an advantage
over internal cyclones in that there is access to the cyclone and
diplegs without going inside the unit. Often there is an extra
layer of erosion-resistant lining in the CFBC cyclones where the
incoming solids first impact the wall of the cyclone.
There is another advantage of using a single large cyclone when
using one cyclone is possible. When using parallel cyclones, the
gas and solids flow into each cyclone is generally not the same
(Whiton, 1941; Smellie, 1942; Koffman, 1953; Broodryk and Shingles,
1995; Grace, 2005; Knowlton et. al., 2016). This unequal flow of
gas and solids manifests itself in different erosion rates, unequal
buildup of coke (and different collection
Fig. 8. Angled Inlet
0.0001 0.001 0.01 0.1 1 10 1000
20
40
60
80
100
Solids Loading, kg /kgs g
Effic
ienc
y, %
Strand SeparationEfficiency
Total SeparationEfficiency
Inner VortexSeparation Efficiency
Fig. 9. Muschelknautz et al., 1996 critical loading plot
-
41
efficiencies) in the parallel cyclones. Whiton, 1941 using small
5-cm diameter cyclones, found that the efficiency of a single
cyclone for his system was 96%, while 7 cyclones in parallel and 14
cyclones in parallel gave lower efficiencies of 94.1 and 92.2%,
respectively.
For large CFBC units, it is also necessary to use parallel
cyclones. One type of parallel arrangement is shown in Fig. 10 for
four cyclones in parallel. Because of differing solids flows
through parallel cyclones, differences have been noted.
Kim et al., 2007 and Stringer and Stallings, 1991 both found
unequal erosion occurring in parallel CFBC cyclones. Hartge et al.,
2005, observed that the temperatures near the bottom of CFBC
parallel cyclone return lines were different. Smellie, 1942 found
that the solids flow rate through parallel cyclones could vary by a
factor of 2. Knowlton et al., 2016 found that with 4 parallel,
300-mm diameter cyclones using 156-micron coke, measured solids
flow rates through parallel cyclones could vary by a factor of
4.
Unlike the axial-inlet cyclone, the tangential and volute
cyclone inlets cause asymmetric flow in the cyclone. As a result,
eddies form near the back of the gas outlet tube (GOT), also called
a vortex finder, in the area approximately 200 to 250 away from the
inlet. For FCC cyclones, this eddy formation leads to coke buildup
on the back side of the GOT. If not addressed, the coke can fall
off and block the dipleg. One method for addressing this problem
has been to add anchors (similar to refractory anchors) on this
area of the GOT wall so that the coke will be held in position as
it forms and not fall off to plug cyclone diplegs.
For CFBC cyclones, one way this asymmetric inefficiency has been
addressed is by changing the position of the GOT. Trefz, 1992 found
that moving the axis of the GOT away from the axis of the cyclone a
short displacement along the radius of the cyclone at a position
approximately 255 away from the inlet to the cyclone, significantly
reduced these eddies, and improved cyclone efficiency. The
resulting GOT offset has been called the eccentric GOT (Fig. 11).
In this figure, the GOT has been moved from the center (dotted
circle) to its final location (continuous line circle) used for the
eccentric GOT. Subsequent implementation of the eccentric GOT in
commercial units has been proven successful (Muschelknautz and
Muschelknautz, 1999). Ipsen et al, 2014 reported that adding an
eccentric GOT (as well as reducing the GOT diameter) to one of the
CFB boilers operated by Stadtwerke Flensburg GmbH improved cyclone
efficiency as well as the operation of the boiler. Before the
improvements, they could not operate at full load. With the
improvements to the cyclone, they could operate at full load, and
the size of the circulating ash in the boiler was reduced by 60
microns improving the heat transfer in the unit and increasing the
amount of circulating ash so that much less sand was required to be
added to the unit to maintain enough circulating solids for
adequate thermal efficiency.
Werther, 2005 reported that using an eccentric cyclone GOT as
well as adding the angled cyclone inlet shown in Fig. 8, improved
the operation of the Ceran B CFB steam generator significantly. The
original Ceran A CFB boiler was not able to prevent the loss of
fine inert material. The Ceran B boiler was an exact duplicate of
the Ceran A boiler except for the cyclone design. Adding the
improved angled inlet and using the eccentric GOT, improved the
cyclone efficiency by such a degree that the median diameter of
the
. The improved cyclone efficiency resulted in finer circulating
solids, an improved heat transfer coefficient and lower NOx and SOx
emissions. Although using an eccentric GOT (and angled inlet) has
proven to be beneficial for CFBCs, the FCC industry has not yet
adopted these improvements.
Kobylecki and Bis, 2008, also showed how increasing cyclone
efficiency reduced the emissions of NOx and SOx. They showed that
increasing the cyclone efficiency from approximately 99.779 to
99.930% reduced the emissions of NOx from approximately 200 to 100
mg/Nm3, and the emissions of SOx from approximately 450 to 200
mg/Nm3.
Eccentric GOT
Fig. 11. Eccentric Gas Outlet Tube
Fig. 10. Typical CFBC configuration for four cyclones in
parallel
-
42
As noted above, increasing the cyclone efficiency in CFBCs can
lead to improved operation. However, sometimes improved cyclone
efficiency is not what is desired. When high ash materials are
being combusted, often the amount of material being circulated is
too much. This can cause a decrease in temperature in the combustor
because of too high of a circulation rate around the system. If the
ash content of fuels in a CFBC varies widely, then it is desired to
have some means of regulating the solids circulation rate.
Muschelknautz and Roper, 2008, developed a method of adjusting the
circulation rate by changing the amount of material lost by the
cyclone.
This was accomplished by adding a gas flow (about 1.3% of the
total flue gas flow) through a nozzle, located on the outside wall
of the cyclone, that was angled/directed to inject solids rotating
on the wall into the vortex just below the GOT (Fig. 12). This
technique reduced the inventory of the bed significantly. The
pressure measured at the bottom of the bed decreased from about 62
to 50 mbar in one test with the nozzle, and from 88 to 55 mbar in
another.
Normally, most cyclones are designed so that the exit velocity
through the GOT is higher than the gas inlet velocity. The high
centrifugal force then throws many of the fine particles not
collected by the outer vortex to the wall, where they are
collected. The higher the outlet gas velocity, the faster the inner
vortex spinning rate and the higher the cyclone efficiency.
Therefore, the inner cyclone vortex is an important element in
cyclone operation.
Because the cyclone inlets in most cyclones are asymmetrical
(with a single inlet), the inner vortex tends to precess or wobble.
The inner vortex also does not taper down to a small tip (as you
would see with a tornado). Visual observation at PSRI in a
Plexiglas cyclone has shown that the inner vortex has a diameter of
approximately 75 to 85% of the diameter of the GOT, and this
diameter is essentially constant throughout the length of the inner
vortex. These observations were made
by operating the cyclone with no solids flowing into it, and
then adding a small amount of fine solids to the cyclone. The fine
solids would escape the cyclone through the GOT, so the shape
of the vortex could be seen.
Foster Wheeler and others have developed what is termed a
compact CFBC (Chen and Jian, 2011). This CFBC uses rectangular or
octagonal solids separators instead of a conventional circular
cyclone. This type of design has several advantages: 1) no
expansion joint is required between the reactor and the solids
separator, 2) the footprint of the unit is reduced, 3) it is
simpler and less costly to build, and 4) the water-cooled flat
panels can be covered with a thin, 5-cm layer of
abrasion-resistance refractory like water-cooled cyclones. The thin
refractory layer allows faster startup and shut down times of CFBC
units. The elimination of the expansion joint also significantly
reduces the down-time of the CFBC - which is a problem with large,
hot refractory-lined cyclones.
A drawing of two CFBs with Foster Wheeler-type compact solids
separators is shown in Fig. 13 from Zhu, 2013. The drawing on the
left shows a rectangular type of solids separator with two vortex
finders, and the drawing on the right shows a
Vortex Finder
Nozzle
Fig. 12. Gas nozzle to regulate solids flow out of a cyclone
Fig. 13. Compact Solids Separators
Outer OctagonalShell
InnerCylinder
VoluteInlet
Fig. 14. Schematic Drawing of an Octagonal Cyclone
-
43
multi-panel solids separator. The corners of the rectangular
cyclone are smoothed with refractory to improve separation
efficiency, but the rectangular solids separators have a lower
collection efficiency compared to a cylindrical cyclone. The
Lagisza CFBC plant in Lagisza, Poland has an octagonal cyclone (Fig
14), which has an improved solids collection efficiency relative to
a rectangular solids separator (Nowak and Mirek, 2013). It would
appear that an octagonal separator is a good compromise between a
cylindrical cyclone and a rectangular solids separator with regard
to solids collection efficiency, cost and ease of construction.
Cyclone Erosion In CFBC processes where only primary cyclones
are used,
most erosion is observed where the solids impinge upon the
cyclone wall near the inlet, and perhaps some erosion will occur on
or in the GOT. In FCC units, most often the critical erosion area
is in the cone of the secondary cyclone. A low-loaded cyclone
(although a non-FCC cyclone) that has undergone erosion in its cone
is shown in Fig. 15.
In an FCC primary cyclone, if the solids flow into the cyclone
is at a rate of 10,000 arbitrary units per hour and the efficiency
of the cyclone is 99.95%, then only 5 units per hour (5/10,000, or
0.05%, of the material flowing into the primary cyclone) will flow
into the inlet of the secondary cyclone (Fig. 16). It seems
counter-intuitive that this low solids flow rate would cause such
severe erosion in the cone of the secondary cyclone. However,
secondary cyclone cone erosion is one of the primary problems in
the shutdown of FCC units.
The reason for the erosion in the FCC secondary cyclone cone can
be seen with the aid of Fig. 17. The primary cyclone has many more
solids flowing through it, but the solids rapidly drop down through
the cyclone after inertial separation from the gas as shown in Fig.
9. Also, the vortex spin rate is decreased because
of the high solids inlet loadings as discussed above, so the
solids are not accelerated at high velocity in the conical region.
It is also likely that the effective vortex length is shortened in
highly-loaded cyclones. For these reasons, there is essentially no
erosion on the primary cyclone cone. The important erosion zone for
primary cyclones is on the wall opposite the solids inlet.
In secondary FCC cyclones, the inner vortex spins at a higher
velocity than in a primary cyclone, and it is also longer. As the
swirling solids move into the cone in a secondary cyclone, the
high-
10,000
9,995
5
4
Efficiency = 99.95%
Solids Gas
Efficiency = 80%
Fig 16. Relative loading in primary and secondary cyclones
Cyclone Loading: HIGH Cyclone Loading: LOW
SlowerVortex
Solids Flow PathMany More Turns: More
Potential for Erosion
Rapidly Rotating Vortex
Fig. 15. Cone Erosion in a low-loaded cyclone
Fig. 17. Solids flow in low- and high-loaded cyclones
-
44
velocity vortex accelerates the solids and increases their
velocity. It is this high-velocity, concentrated-solids stream that
causes the significant erosion seen in secondary cyclone cones
(Fig. 15) and in the upper part of the secondary cyclone
diplegs.
There are several things that can be done to mitigate this
erosion: 1) increase the length of the cyclone (cone, barrel or
both), 2) add a dust hopper, or 3) add a vortex stabilizer. A
cyclone with one common type of dust hopper is shown in Fig. 18. Of
these different solutions, it was found that the most effective
method of reducing the secondary cyclone cone erosion was to add a
vortex stabilizer in the bottom part of the cone of the cyclone.
Shell has used the vortex stabilizer for over 20 years to prevent
excessive erosion in their secondary cyclone cones (Chen, et al.,
2013).
The vortex stabilizer minimizes the precessing of the inner
cyclone vortex, but primarily prevents the vortex from extending
below the vortex stabilizer. If the vortex stabilizer is placed in
the cone a certain distance away from the bottom of the cone, the
rapid swirl of the inner vortex is reduced to a very low velocity
below the vortex stabilizer, and the cyclone erosion produced at
the bottom of the cone is significantly reduced. The larger the
diameter of the vortex stabilizer, the more effective it is in
reducing erosion. A plot showing the relative effectiveness of
increasing the vortex stabilizer diameter for a flat plate cyclone
vortex stabilizer is shown in Fig 19. Superimposed on the plot are
data points
showing how increasing cyclone barrel length, increasing cone
length or adding a dust hopper reduces the secondary cyclone cone
erosion rate. All of these methods reduce the erosion rate in the
secondary cyclone cone, but adding a vortex stabilizer in the
cyclone cone region is significantly more effective than the other
methods. The effectiveness of using a flat plate vortex stabilizer
relative to having no vortex stabilizer is shown in Fig. 20. In
this figure, the erosion rate on the secondary cyclone cone is
shown as a function of the gas velocity in the
GOT, at a constant inlet velocity in the cyclone of 19.8 m/s and
at a constant solids loading of 0.032 kg/m3. The velocity in the
GOT was varied by changing the diameter of the GOT. This figure
shows that the cyclone cone erosion was significantly reduced using
the flat-plate vortex stabilizer vs. no vortex stabilizer as the
velocity in the gas outlet tube was increased. It was also found
that the insertion of the vortex stabilizer into the bottom of the
cone did not increase nor decrease the collection efficiency of the
cyclone.
The erosion rates shown in Figs. 19 and 20 were measured by
coating a Plexiglas cone with drywall joint compound and weighing
the amount of drywall joint compound lost to
Dust Hopper
Fig. 18. Cyclone with Dust Hopper
Fig 19. Effect of Flat Plate Vortex Stabilizer Diameter on
Erosion Rate
Fig.20. Cone Erosion Rate vs. gas velocity in the GOT for a
flat-plate vortex stabilizer
-
45
erosion after a test. Specifically, three different layers of
the compound were added to the cone, and the weight of the cone and
drywall joint compound measured. After the cyclone was operated in
a recirculating solids system for several minutes, the cyclone was
then removed from the system and the cone and remaining drywall
joint compound weighed to determine the erosion rate (Chen, et al.,
2013).
Several types of vortex stabilizer configurations can be used.
The flat plate, cone and spike type of vortex stabilizer
configurations are shown in Fig. 21. Equivalent base diameters of
these different configurations appear to give the same
effectiveness in reducing erosion. As shown in Fig. 19, the larger
the diameter of the vortex stabilizer, the more effective it is in
reducing erosion. For best operation of the vortex separator, the
diameter of the bottom of the vortex separator should at least be
equal to the diameter of the inner vortex, which is approximately
75 to 85% of the diameter of the GOT, as discussed above. However,
it should not be made so large in diameter that
it affects the flow of solids down the wall of the cone.
Surprisingly, the top of the vortex stabilizer does not erode, as
minimal erosion was observed on the top surface of each type of
vortex stabilizer. When supporting the vortex stabilizer from the
wall of the cone, the supports should be below the vortex
stabilizer. Adding supports above the base of the vortex stabilizer
will result in erosion of the supports.
The vortex stabilizer has another advantage when it is used in
FCC secondary cyclones. The vortex stabilizer modifies the pressure
profile in the cyclone so that the pressure at the solids discharge
of the cyclone is higher than without the vortex stabilizer
(Muschelknautz and Grief, 1997). This means that the required
pressure build in the dipleg below a cyclone containing a vortex
stabilizer will be less than in a dipleg below a cyclone without a
vortex stabilizer, requiring shorter diplegs.
The pressure drop across CFBC primary cyclones can be
significantly reduced by adding swirl vanes into the outlet of the
GOT (Ipsen et al., 2014). Adding swirl vanes to the outlet of a
commercial CFBC reduced the pressure drop by approximately 30 to
40% for cyclones with inlet loadings above 0.5 kg/m3, and by up to
60% for inlet cyclone loadings below 0.1 kg/m3
Another approach to reducing the pressure drop across solid
separators in a CFBC is to not use cyclones at all. B&W uses
U-beams (Fig. 22) in their CFBCs to separate the solids from the
exit gas (Maryamchik and Wietzke, 2010). The collection efficiency
of the U-beams (generally between about 95 and 97%) is not as high
as for a cyclone (which is about 99.9% or a little more) or the
compact separators (perhaps 98 to 99%). Therefore, a multiclone
type of cyclone is generally used as a second stage of solids
collection. U-beams are used because of their low pressure drop.
Pressure drops as low as 1kPa have been reported with U-beams.
Cyclone Diplegs and Dipleg Terminations FCC primary cyclone
efficiencies can achieve cyclone efficiencies of 99.95 to 99.99%.
Secondary FCC cyclones can achieve cyclone efficiencies of 98 to
99%, if the secondary cyclone dipleg is designed and operated
correctly. In all processes utilizing a cyclone, the cyclone should
be thought of as part of a system, and not separately. The
operation of a cyclone will depend on how the rest of the system
(especially the dipleg) is designed. If the secondary cyclone
dipleg in FCC units is not designed correctly, too much gas can
flow up the dipleg and reduce the efficiency of the cyclone
significantly. In CFBC cyclones, it is also possible to have
U-Beam
Gas + SolidsGas
Fig. 22. Top View of U-beams
Fig. 21 Types of Vortex Stabilizers
Flat PlateVortex Stabilizer
Conical
Spike
Support
-
46
too much gas flow up the dipleg and reduce the efficiency of the
primary cyclone (Muschelknautz and Grief, 1997). For cyclones
operating in FCC units, the secondary cyclone dipleg is required to
have a device such as a trickle vale or counterweighted flapper
valve (Fig. 23) at the end of the dipleg. Primary FCC cyclone
diplegs do not require trickle valves or flapper valves because of
the high solids fluxes in their diplegs. However, some FCC primary
diplegs have trickle valves on them to try to prevent catalyst loss
upon startup.
A trickle valve or counterweighted flapper valve is required at
the end of secondary cyclone diplegs at the startup of the FCC
unit, which starts up with the fluidized bed empty. As the solids
are added to the FCC unit, the solids pass through the primary
cyclone. As shown in Fig. 16, for 10,000 arbitrary mass units of
solids entering the cyclone per unit time (assume the units are
kg/min), only about 5 kg/min will enter the secondary cyclone. If
the efficiency of the secondary cyclone is 80%, then only 4 kg/min
will flow down the secondary cyclone dipleg.
Fluidizing gas is being added to the vessel below the FCC
cyclone during startup. The solids flux down the primary cyclone
dipleg (generally designed to have a solids mass flux of between
about 500 to 750 kg/m2/s) is so great that the gas cannot flow
up
the primary dipleg against this large flux, and will be carried
down the primary dipleg with the solids (Karri and Knowlton, 2001).
However, for the case illustrated in Fig. 16, the solids flux down
the secondary cyclone dipleg will only be 4/10,000 of the primary
cyclone dipleg flux (if the same dipleg diameter as the primary
dipleg is used), or only 0.2 to 0.3 kg/s/m2. This is 2500 times
less than the flux in the primary dipleg. Using a smaller-diameter
dipleg to try to increase this flux to values similar to those in
the primary dipleg leads to impractically-small, secondary cyclone
diplegs.
This extremely low flux of solids down the secondary dipleg is
not enough to prevent gas from flowing up the secondary dipleg in
amounts great enough to prevent solids flow down the secondary
dipleg. This large gas flow upward prevents a solid seal from
developing in the secondary dipleg. Therefore, a device such as a
trickle valve or a counterweighted flapper valve is added to the
end of the secondary cyclone dipleg to prevent most of the gas from
flowing up the dipleg. This allows a solids seal to develop in the
dipleg. However, even with the trickle valve or counterweighted
flapper valve at the bottom of the dipleg, gas will still flow up
the secondary dipleg as shown in Fig. 24, Karri and Knowlton,
2001.
If the FCC secondary cyclone dipleg is not designed and
operated
correctly, a significant amount of gas can flow up the dipleg
and reduce the efficiency of the secondary cyclone and increase the
emissions from the FCC unit. PSRI has found that if the upward gas
velocity in the FCC secondary cyclone dipleg is greater than about
1 m/s, then the cyclone efficiency can decrease from about 98 to
99% to about 75 to 80% (Karri, 2015). This effect is shown in Fig.
25, and is similar in effect to what was found above for excessive
upward dipleg gas flow affecting cyclone efficiency for CFBC
cyclones. Many FCC secondary cyclone diplegs have their solids
discharging into the dilute-phase freeboard above the
Fig. 23. Schematic Drawing of Trickle and Counterweighted
Flapper Valves
0 100 200 300 400 500 600 7000
0.5
1
1.5
2
2.5
3
3.5
Solids Mass Flux in Dipleg, kg/s-m2
Cyc
lone
Inle
t Gas
Flo
win
g U
p/D
own
Dip
leg,
%
Material: 76-micron FCC CatalystDipleg Dia: 10 cmDipleg Length:
2.5 mCyclone Inlet Gas Velocity: 19.8 m/sDipleg Pressure Drop: 4.3
kPa
UP DOWN
Positive Cyclone
Fig. 24. The Effect of Solids Mass Flux in the Dipleg on the
Direction of Gas Flow for FCC Catalyst
-
47
fluidized bed. Therefore, the tolerances between the trickle
valve (or counterweighted flapper valve) plate and the dipleg
should be as small as possible to prevent an excessive amount of
gas from flowing up the dipleg and reducing the secondary cyclone
efficiency. However, a better way of reducing the amount of gas
flowing up the dipleg is to immerse the secondary dipleg into the
fluidized bed (Fig. 26). This increases the height of the solids
seal required in the dipleg, increases the resistance to gas flow
up the dipleg, and significantly reduces the gas flow (and the
velocity) up the dipleg. Immersions of about 1 m into the fluidized
appear to be satisfactory to reduce this gas flow.
However, if immersions are too deep, then the solids seal height
will be too close to the bottom of the cyclone.
CFBC diplegs can also have gas flowing up the dipleg, even with
relatively high solids fluxes. The solids in the dipleg of the CFBC
cyclone dipleg are larger (approximately 120 to 200 microns vs.
about 65 to 70 microns) than in the FCC primary cyclone dipleg.
Tracer studies at PSRI with 120-micron sand material flowing
through a dipleg above a loop seal have shown that for solids mass
fluxes less than about 300 kg/s/m2 through the dipleg, gas flowed
up the dipleg. This is a much higher mass flux than the limiting
flux shown in Fig. 24 (about 75 kg/s/m2) required for gas to flow
up the dipleg for FCC catalyst. The reason is that the surface
area-per-unit-volume of the larger sand is much lower than that of
the smaller FCC catalyst, so it takes a much higher mass flux
(velocity) in the line to carry the gas down with the solids. The
sand also has a greater flowing density in the dipleg than the FCC
catalyst (approximately 1500 kg/m3 for the sand vs. about 750 kg/m3
for the FCC catalyst). The velocity in the dipleg where gas flow in
the FCC catalyst suspension approximately transitions from up to
down in the dipleg is 75 kg/s/m2/750 kg/m3 = 0.1 m/s. The velocity
in the sand dipleg where the solids carried the gas down the dipleg
was about 300 kg/s/m2/1500 kg/m3 = 0.2 m/s. Because of the
difference in particle size and particle density, comparing
velocities is a more accurate way of comparing differences in
dipleg operation. However, this analysis shows that gas can travel
up the dipleg at higher gas/solids suspension velocities in CFBC
diplegs than in FCC diplegs, primarily because of the difference in
the particle sizes in the two processes. The higher upward gas
velocities in the dipleg below CFBC cyclones can also affect CFBC
cyclone efficiency as (Muschelknautz and Grief, 1997) indicated
above. However, the amount of gas flow up the dipleg below CFBC
cyclones is also largely a matter of how much the loop seal is
aerated. Unless the flux in the dipleg is high enough to prevent
upward gas flow, if the loop seal is overaerated it can affect not
only the operation of the loop seal, but the efficiency of the
cyclone above the loop seal as well.
In FCC units, a recurring problem is the loss of fine material
from the process. Nearly always, this is not the fault of the
cyclone itself, but a problem with the secondary cyclone dipleg. In
this dipleg, the solid flux is so low, and the particle size so
small, that solids flow through this dipleg can be blocked, or so
much gas can flow up the secondary cyclone dipleg (because of the
reasons discussed above) that a significant amount of material can
be lost. Often the problem is with the trickle or flapper valve
plate not sealing well (because of poor design or warping), causing
too much gas to flow up the dipleg which affects cyclone efficiency
as indicated above. As also indicated, immersing the trickle or
flapper valve in the bed a short distance can mitigate this
problem.
Normally, a trickle or a flapper valve on the outlet of a
secondary cyclone dipleg operates in an intermittent discharge
mode. Solids will build up into the dipleg to a height which will
provide enough head to generate enough force to open the flapper
plate. Often there will be a continuous, small trickling flow of
the solids through the trickle valve or flapper valve which
constitutes
1 m
A B
TrickleValves
Fig. 26. Immersed and Non-immersed Trickle Valves
Fig. 25. Cyclone Efficiency vs. Gas flow Up the Dipleg
70
75
80
85
90
95
100
0 0.2 0 .4 0 .6 0 .8 1 1.2 1 .4 1 .6
Cyclo
ne E
fficie
ncy,
%
Gas Velocity Up Dipleg, m/s
Dipleg Diameter: 10 cmGas: AirLoading: 0.019 kg/m3Material: FCC
CatalystUin: 19.8 m/sUout: 27 m/s
-
48
approximately 10% of the solids flow rate. (Geldart and
Kerdoncuff, 1992). Superimposed upon this constant trickling is the
intermittent discharge mode, which discharges the other 90% of the
solids flow rate. At times, the solids flow rate is so low, that
the solids in the secondary dipleg can defluidize (deaerate) before
the solids reach the height necessary to produce the pressure
required to open the flapper on the trickle or flapper valve. If
this occurs, then the defluidized bed of solids above the valve
cannot produce enough head to open the flapper of the valve, and
solids flow out of the dipleg will stop. If this occurs, the dipleg
will flood, which means that the dipleg will fill up with solids to
the bottom of the cyclone. When this occurs, the solids flowing
into the cyclone will bypass out through the cyclone exit.
One way to solve this problem having defluidized solids in the
dipleg is to add aeration into the dipleg. The best location to add
aeration for the trickle valve is to add it in the mitered section
at the bottom on the centerline of the dipleg Karri and Knowlton
(2004). This location is shown in Fig. 27. They also found that
adding aeration such that the superficial gas velocity in the
dipleg was around 0.03 m/s gave good results.
However, adding aeration to a dipleg in a hot unit is not easy
to do. There are usually between about 6 and 22 sets of cyclones in
an FCC unit, and adding aeration lines to all of their secondary
diplegs with the associated thermal expansion problems, is
complicated. Therefore, it is easier to reduce the size of the
cyclone dipleg so that the solids level will reach the required
height to open the flapper before deaeration occurs.
Loop seals at the bottom of the cyclone dipleg in CFBCs are very
reliable. For a loop seal to operate correctly, the upleg of the
loop seal must be fluidized (Fig. 28). The downleg of the loop seal
should also be fluidized for best operation, although the loop seal
can operate with a non-fluidized downleg. However, it is strongly
recommended that the downleg be fluidized for best operation of the
loop seal.
In CFBC processes, the loop seal is the most extensively-used
device to return solids collected by the cyclone back into the
reactor. The FCC process does not make use of loop seals thus far
primarily because the cyclone diplegs are internal to the process.
Loop seals can be used with both Geldart Group A or Geldart Group B
solids. However, it would be difficult to start up an FCC unit with
an empty loop seal instead of a
trickle valve or flapper valve. The loop seal would have to be
filled with solids before startup, so that the gas would not flow
up the secondary dipleg and prevent a solids seal from forming due
to the low secondary cyclone dipleg solids flow as indicated in
Fig. 16.
Attrition in Cyclones One issue associated with cyclones is
particle attrition. The solids enter the primary cyclone at a high
velocity and impinge on the wall of the cyclone. This can cause
substantial particle attrition. In FCC units, cyclone attrition is
almost always of the surface abrasion type where the small surface
nodules and asperities are abraded off of the surface of the FCC
catalyst. These surface asperities are small, of the order of about
0 to 10 microns, and are mostly lost into the cyclone exit stream
as the FCC cyclones have poor collection efficiencies for these
sizes of particles.
In CFBC systems, attrition can be due to thermal stress,
chemical reaction (which weakens the particles as they are
reacted), as well as mechanical attrition. Mechanical attrition is
what occurs in the cyclones, but the thermal shock or chemical
reaction can weaken the particles so that they can more easily be
broken.
Scala and Chirone, 2013, found that when a CFBC cyclone inlet
gas velocity exceeded a certain threshold velocity, that the
attrition produced by the cyclone increased significantly. They
attributed this sudden increase to a chipping of material off of
the particle surface (a type of fragmentation). However, the
authors also noted that for typical cyclone inlet velocities found
in CFBC cyclones, this type of fragmentation would not occur, and
the primary mode of attrition would be particle
Fig. 27. Optimum Aeration Location for a
Trickle Valve
DownlegUpleg
Aeration
Fig. 28. Schematic Drawing of a Loop Seal
-
49
abrasion. This is because high cyclone velocities cause the
cyclone pressure drop to increase to values higher than desired, as
cyclone pressure drop is proportional to inlet gas velocity
squared.
Reppenhagen and Werther, 1999 developed a correlation for
attrition in a cyclone based on their work with FCC catalyst. They
found that the attrition in their 90-mm ID cyclone using this
catalyst was essentially surface abrasion. The correlation they
developed for this type of attrition is:
This correlation predicts that the attrition rate in a cyclone
is proportional to the inlet velocity squared, and inversely
proportional to the square root of the mass loading (kgs/kgg)
entering the cyclone. In catalytic FCC systems, this large
dependency of attrition on inlet gas velocity has been recognized
and used to reduce particle attrition in cyclones. Because particle
emission requirements are becoming more and more stringent, and a
significant contribution to the emissions is material attrited in
the cyclones, in the authors experience, there is a growing trend
toward designing primary FCC and other catalytic process cyclones
with lower inlet gas velocities. This is because of the strong
dependency of attrition on gas inlet velocity.
As can be seen from Fig 16, the loading is so high in the
primary cyclone relative to the secondary cyclone that,
practically, all attrition occurs in the primary cyclone. The
solids flow rate into the secondary cyclone is typically between
1/1,000 to 1/10,00 of the flow rate into the primary cyclone, so
relatively little attrition occurs in the secondary cyclone.
Typical primary cyclone inlet gas velocities for FCC units have
traditionally been in the range of 18.3 to 19.8 m/s (60 to 65
ft/s). Several primary cyclones have now been designed to operate
in the range of 13.7 to 16.8 m/s (45 to 55 ft/s). Reducing the
inlet gas velocity from
18.3 to 15.2 m/s, will reduce particle attrition caused by the
cyclone by approximately 31% if cyclone attrition is proportional
to inlet gas velocity squared. However, reducing the inlet gas
velocity will also increase the size of the cyclone, or may require
increasing the number of cyclones. Using the same relative
dimensions of the cyclone with a reduced inlet gas velocity, the
cyclone would have to be increased in diameter by approximately
10%. As well as reducing the attrition produced in the primary
cyclone, the erosion of the cyclone impact area would be reduced as
well by the reduction in the inlet gas velocity. To counteract the
effect of the lowering of the inlet gas velocity, the GOT diameter
can also be reduced to increase the rotational speed of the inner
vortex, and increase the cyclone collection efficiency.
Agglomeration of fine solids can also affect cyclone efficiency
for these very fine solids (generally smaller than 10 microns). In
general, agglomerate of these fine particles result in an increase
in cyclone efficiency. The fine particles clump together
(agglomerate) because of interparticle forces. Evidence of this
clumping can be seen in Fig. 29. In this figure, very fine TiO2
solids with an average particle size of approximately 2.2 microns,
were directed into a
Fig. 29. Cyclone Efficiency vs. Inlet Gas Velocity for TiO2
2 3 5 10 20 30 50 10090
92
94
96
98
100
Particle Diameter, microns
Frac
tiona
l Effi
cien
cy, %
Material: Glass Beads (40 microns)Loading: 40 kg solid/kg
gasInlet Gas Velocity: 12.4 m/sCyclone Diameter: 63 cm
Fig. 30. Cyclone Efficiency vs. Particle Size With Indication of
Small-Particle Clumping
86
87
88
89
90
91
92
93
94
6 8 10 12 14 16 18 20
Cyclo
ne E
fficie
ncy,
%
Inlet Gas Velocity, m/s
Material: TiO 2dp: 2.2 micronsSolids Loading: 0.32 kg/m 3Cyclone
Dia: 43 cmInlet Type: TangentialGas: Air
-
50
tangential inlet cyclone at several gas velocities ranging from
approximately 8.2 to 18.2 m/s. The gas velocity through the cyclone
was varied at a constant inlet loading of 0.32 kg/m3. A very low
cyclone collection efficiency was expected for these particles, as
cyclones generally cannot collect solids of this size at a high
efficiency. However, a very high collection efficiency (over 92 %)
was obtained for these very small particles at an inlet gas
velocity of 8.2 m/s. As the inlet gas velocity was increased, the
collection efficiency for these particles decreased. The reason for
this counter-intuitive result appeared to be that the small TiO2
particles clumped together in hydrodynamically stable, but
relatively fragile agglomerates. As the gas velocity was increased,
the increased impact force of the agglomerates with the cyclone
wall caused an increased percentage of the agglomerates to break up
into smaller particles, which could not be collected by the
cyclone. Therefore, the cyclone efficiency decreased as the gas
velocity increased.
The effect of particle clumping at small particle sizes can also
be seen in Fig. 30. In this figure, the fractional collection
efficiency of glass bead particles is plotted as a function of
particle size (from Hugi and Reh, 1998). This fractional efficiency
plot shows that the collection efficiency of 2-micron particles is
greater than or equal to the collection efficiency of 20- to
30-micron particles. This is because the smaller particles clump
together to form larger agglomerate particles which can be
collected at higher efficiencies than larger particles that do not
agglomerate (at least to the same extent).
Positive and Negative-Pressure Cyclones A negative-pressure
cyclone is a cyclone where the pressure in the barrel is less than
the pressure in the freeboard of the bed into which it is
discharging. A positive-pressure cyclone is a cyclone where the
pressure in the barrel is greater than the pressure of freeboard of
the bed into which it is discharging.
All CFBC cyclones are negative-pressure cyclones. However, some
FCC cyclones can be either negative -pressure or positive-pressure
cyclones. FCC close-coupled cyclones at the exit of the FCC riser
can be either negative-pressure or positive-pressure cyclones.
These cyclones are generally located above the stripper section of
the FCC unit. Whether they are positive-pressure or
negative-pressure cyclones depends on how the gas from the stripper
enters the cyclone system. If the gas enters before the inlet of
the primary cyclone, the cyclone is a negative-pressure cyclone. If
the gas from the stripper enters between the primary and secondary
cyclones, the cyclone is generally a positive-pressure cyclone.
This difference is shown in Fig. 31. The secondary cyclones are not
shown in this figure.
As shown in Fig. 31, if the cyclone is a positive-pressure
cyclone, the level of the solids seal in the dipleg will be below
the level of the stripper bed. If the cyclone is a
negative-pressure cyclone, the level of the solids seal in the
dipleg will be above the level of the stripper bed.
A positive-pressure cyclone is also more likely to have a dipleg
that operates in what is called the streaming mode (Karri and
Knowlton, 2001; Dries and Bouma, 1996; Sun, et al., 2001). In this
mode, the solids stream down the primary cyclone dipleg and drag a
significant amount of gas with them. Fortunately, this not a common
occurrence, as a dipleg operating in the streaming mode is almost
always non-desirable as it increases the load on the stripper. The
dipleg streaming mode is more likely to occur if the pressure drop
across the primary dipleg is low. The pressure drop across the
primary dipleg for a positive-pressure cyclone can be significantly
lower than the pressure drop across the primary dipleg for a
negative-pressure cyclone, as shown in Fig. 31.
SUMMARY The two most important CFB industrial processes that use
cyclones are the FCC and the CFBC processes. The cyclones from
these two processes have important differences. One difference is
in the size of the solids which pass through the cyclones. The FCC
process uses a catalyst which is a Geldart Group A material (about
70
Riser Gas
P = 21
Pressures are in kPag
P = 7
P = 4.2 P = 9.1
P = 16.8 P = 11.9
NegativePressureCyclone
PositivePressureCyclone
Riser Gas
Stripper Bed
Gas In
Gas In
Fig. 31. Positive and Negative Cyclones
-
51
microns in diameter), and the CFBC process uses a larger, Group
B material (typically 150 to 200 microns in diameter).
The size and configuration of the cyclones in the two systems
are also different. FCC cyclones are approximately 1.2 to 2 m in
diameter, are mostly parallel, internal cyclones that have at least
two and can have as many as four stages. CFBC cyclones are much
larger (up to 10 m in diameter), are mostly external cyclones, and
have only one stage.
The efficiencies of the multi-stage FCC cyclone system can be as
much as 99.999%, whereas the larger, single stage CFBC cyclone
efficiencies are more typically about 99.9%. CFBC systems which use
square or multi-panel non-cylindrical cyclones for lower costs and
reduced maintenance have efficiencies somewhat less than this.
CFBC units almost always use a loop seal at the bottom of their
external cyclone dipleg to seal the differential pressure between
the cyclone and the reactor. This type of dipleg return system is
not used in FCC units. FCC units use either trickle valves or
flapper valves at the bottom of their secondary cyclone diplegs to
allow the units to startup with an empty bed.
Most erosion in CFBC cyclones occurs in the cyclone barrel
opposite the point where the solids enter the cyclone. The most
severe cyclone erosion in FCC cyclones, occurs in the cone of
secondary cyclones, and adding a vortex stabilizer to the appears
to be the best way to mitigate this erosion.
For CFBC cyclones, it has been found that using an eccentric GOT
and an angled inlet can both improve the collection efficiency of
the CFBC cyclones. These configurations have not been tried in FCC
cyclones.
Both types of cyclones can have their collection efficiencies
affected if there is too much gas that flows up the cyclone dipleg.
This primarily occurs in the secondary cyclone diplegs for FCC
cyclones, and for lower-flux CFBC cyclone diplegs.
NOTATIONCc attrition rate constant, s2/m3 dpc surface mean
particle size, m D cyclone diameter, m L cyclone overall length, m
kgg kilograms of gas kgs kilograms of solids
rc attrition rate, fines mass rate produced in the cyclone/fines
mass rate entering cyclone, (-) Ui, Uin cyclone inlet gas velocity,
m/s Uo, Uout cyclone outlet gas velocity, m/s Uc cyclone inlet gas
velocity, m/s voidage, (-) c solids to gas mass loading ratio,
kgs/kgg
REFERENCES Broodryk J., Shingles T. 1995. Aspects of cyclone
operation in industrial chemical reactors. Fluidization VIII.
Tours, Eds. Large, J. France. pp. 1083-1090. Burdett, I.,
Eisinger, R., Cai, P., Lee, H. 2001. Gas-phase fluidization
technology for the production of
polyethylene, in: Kwauk, M., Li, J., Yang, W-C. (Eds.),
Fluidization X, United Engineering Foundation, New York, pp.
39-52.
Chen, W., Jiang, P. 2011. Design and operation of CFB with
compact separator. PowerGen Asia 2011 Conference Proceedings, Kuala
Lumpur, Malaysia, p. 9.
Chen, Y., Karri, R., Knowlton, T. 2010. Winning in the downturn:
how to improve FCC unit reliability and reduce costs via improved
cyclone technology, NPRA Meeting, Phoenix, AZ.
Dries, H., Bouma, J. 1996. Down flow of Class A powder in
cyclone diplegs. Circulating Fluidized Bed Technology V, Kwauk, M.,
Li, J., eds. Engineering Foundation, New York. Pp.161-166.
Finch, J. Dust collector. US Patent 325,521. September 1, 1885.
Geldart, D., Kerdoncuff, A. 1992. The behaviour of secondary and
tertiary stage cyclone diplegs. Paper
presented at the 1992 Annual AIChE Meeting in Miami, Florida,
USA. Grace, J. 2008. Maldistribution of flow through parallel
cyclones in circulating fluidized beds, in Circulating
Fluidized Ted Technology IX, (Werther, J. Nowak, N., Wirth,
K.-E. and Hartge, E.-U., eds), TuTech Innovation GmbH, Hamburg,
pp.969-977.
Hartge, E.-U., Budinger, S., Werther, J. 2005. in Circulating
Fluidized Bed Technology VIII (Kefa, C., ed.), International
Academic Publishers, World Publishing Corp., Beijing, p. 675
Hoffman AC, de Jonge R, Arends H, Hanrats C. Evidence of the
Natural Vortex Length and its Effect on the Separation Efficiency
of Gas Cyclones. Filt. Sep. 32:799. 1995.
-
52
Hugi, E., Reh, L. 1998. Design of cyclones with high entrance
loadings. Chemical Technology, 9 ,pp. 716-719.
Ipsen, C., Roschek, D., Muschelknautz, U. 2014. Theory and
operating data of cyclone improvement in a CFB boiler. Proceedings
of the 11th International Conference on Fluidized Bed Technology,
(Li.j., Wei, F., Bao, X., Want, W. eds.). Chemical Industry Press.
Beijing, China, pp.929-934.
Karri, R., Knowlton, T. 2004. The effect of aeration on the
operation of cyclone diplegs fitted with trickle valves. I & EC
Research, Vol. 43, 18.
Karri, R. 2008. Personal communication. Karri, R. 2015. Personal
communication. Karri, S., Knowlton, T. 2001. Streaming flow in
cyclone diplegs. Fluidization X, Kwauk, M., Li, J., Yang,
W., eds. United Engineering Foundation, New York, pp. 731-738.
Kim, T., Choi, J., Shun, D., Kim, S., Kim, S., Grace, J., 2007.
Powder Technology, 178 (2007) p. 143. Knowlton, T., Karri, R. 2008.
Differences in cyclone operation at low and high solids loading.
Proceedings,
Industrial Fluidization South Africa. The Southern African
Institute of Mining and Metallurgy, Johannesburg, pp.119-160.
Knowlton, T., Findlay J., Hackman, L., McKnight, C. 2016. Solids
maldistribution in parallel cyclones. Presented at Fluidization XV,
Eds. Chaouki, J., Bi, X, Berruti, F., Cocco, R., Montobello,
Canada,
Koffman, L. 1953. Gas and Oil Power 48, pp. 89-94. Maryamchik,
M., Wietzke, D. 2010. B&W IR-CFB: Development, projects and
experience. Paper presented
at: Coal-Gen 2010 Conference and Exhibition. Pittsburg, PA, USA.
Muschelknautz E., Krambrock. W. 1996. Pressure drop and separation
efficiency in cyclones. VDI Heat Atlas,
6, and 1. English edition, chap. Lj, 1-8. Muschelknautz, E.,
Grief, V., 1997. Cyclones and other gas-solids separators. Chapter
6 in Circulating
Fluidized Beds, edited by Grace, J., Avidan, A, Knowlton, T.,
Blackie Academic & Professional, London, pp. 181-213.
Muschelknautz, U. and Muschelknautz, E., 1999. Separation
efficiency of recirculating cyclones in CFB combustion. VGB
PowerTech/4/1999, pp. 48-53.
Muschelknautz, U. Roper, B. 2008. On-line regulation of the
solids circulation rate in a circulating fluidized bed combustor.
Circulating Fluidized Bed Technology IX, Werther, J., Nowak, W.,
Wirth, K-E., Hartge, E-U., eds. TuTech Innovation GmbH, Hamburg,
pp.1033-1037.
Muschelknautz, U. 2010. Cyclones for the precipitation of solid
particles. VDI Heat Atlas, 2nd English Edition, Springer, Berlin,
Chapter L3.4, pp. 1-12.
Nowak, W., and Mirek, P. 2013. Circulating fluidized bed
combustion (CFBC). Chapter 16 in Fluidized bed technologies for
near-zero emission combustion and gasification, Scala, F. ed.
Woodhead Publishing Ltd. Pp. 701-764.
Reppenhagen, J, Werther. 1999. Catalyst attrition in cyclones.
Powder Technology 113, pp. 55-69. Scala, F., Chirone, R. 2013.
Attrition phenomena relevant to fluidized bed combustion and
gasification
systems. Chapter 6 in Fluidized bed technologies for near-zero
emission combustion and gasification, ed. Scala, F. pp.
254-315.
Smellie J.. 1942. Iron and Coal Trades Review, 144, p. 169.
Stringer, J., Stallings, J. 1991. Proceedings of the 11th
International Fluidized Combustion Conference,
(Anthony, E., ed.), ASME, New York, 2, p.589. Sun, G., Shi, M.,
Shi, A., Chen, B., Zia, L. 2001. Flow pattern of particles in a
cyclone dipleg with a trickle
valve. Fluidization X., Kwauk, M., Li, J., Yang, W. eds. United
Engineering Foundation, New York. pp. 723-729.
Trefz, M. 1992.Boundary layer flows in cyclones and their
influence on the solid separation. 2. European Symposium Separation
of Solids from Gases, Nuremburg.
Weaver, E., Geiger, F. 2002. FCCU particulate emissions control
with a Shell third stage separator a case study. Proceedings,
National Petrochemical Refining Association Conference, New
Orleans, La, USA.
Werther, J. 2005. Fluid dynamics, temperature and concentration
fields in large-scale CFB combustors. Circulating Fluidized Bed
Technology VII. Cen, K. ed. International Academic Publishers/World
Publishing Corporation. Beijing, China. pp.1-25.
Whiton L., Jr. 1941. Trans. Am. Soc. Mech. Engrs., 213-8. Yuu,
S., Tomosada J., Yuji T., Yoshida, K. 1978. The reduction of
pressure drop due to dust loading in a
conventional cyclone. Chemical Engineering Science, 33, p.
1573-1580. Zhu, Q. 2013. Developments in circulating fluidized bed
combustion. CCC/219, ISBN 978-92-9029-539-6,
IEA Clean Coal Centre, London, pp. 1-60.